These are the droids you’re looking for: An optimistic vision for artificial intelligence, automation and the future of work
The Adam Smith Institute’s latest paper, by Fellow James Lawson, makes the optimistic case for the future of artificial intelligence and employment.
Artificial intelligence
Machine learning is the most important area advancing artificial intelligence (AI). It allows more complex problems to be solved than traditional coding and work to be automated more easily.
AI is real and increasingly used all around us in a wide range of applications, from entertainment to transport, healthcare and office work.
AI’s impact on jobs
There have been widespread concerns about the impact of AI on jobs even before the economic crisis caused by COVID-19. These concerns will only intensify in the challenging period ahead.
There have been similar concerns about the impact of new technology on jobs for centuries.
These worries are often driven by the Luddite fallacy: assuming that robots and workers are competing for a fixed number of jobs in a static economy.
Automation has historically been a force for good and doomsday scenarios have not transpired.
Some jobs are highly vulnerable to automation from AI. An estimated 30-40 per cent of UK work is at high risk.
These studies provide a useful directional guide about the scope of automation but are subjective, may not translate into actual job losses, and are unclear on timelines or the net impact on employment.
The net impact of AI on jobs and the flexibility of the labour market will determine the future outcome. This paper uses a Technological Unemployment Matrix as a framework to guide policymakers.
The most likely scenario is that AI will support greater prosperity. There is no trend so far towards the doomsday scenarios, and the UK labour market is flexible, with a strong record of delivering high employment. AI will create new jobs, boost productivity and increase purchasing power. There will be some losses to mitigate, with temporary displacement and pockets of unemployment.
Government policies
Surveying twenty-five governments’ policy shows a common blueprint: announcing AI leadership intentions through to publishing an AI strategy and pledging funds for research. These promises are shallow, will have little impact and are unlikely to withstand lobbying.
Technology progresses faster than regulation, creating a “pacing problem”. A regulatory vacuum hinders progress. Estonia’s approach, focusing on creating a permissive regulatory environment in which AI can flourish, is instructive.
Potential policy implications
Vague pronouncements and half measures will not position the UK to lead in AI nor mitigate the potential jobs impact. The UK needs a joined-up and radical programme extending across regulation, research and development, welfare and taxation.
Technology underpins economic growth. Government should recognise AI’s potential contribution and the importance of fast adoption to improve the UK’s competitive position.
The UK should adopt a “permissionless innovation” regulatory approach for AI leadership. This contrasts with the default government stance of the “precautionary principle”.
Government should set up a £5 million ‘Office for removing barriers to Artificial Intelligence’ (ORBI) and pass an ‘Unleashing Artificial Intelligence Act’ (UAI Act). The Office would remove impediments to artificial intelligence and make permissionless innovation the legal default. This approach could be expanded to other areas of regulation.
Government should not resort to prohibitions, fines, threats, or licensing except in the extreme and with an understanding of the risks. When intervention is genuinely justified, it should support its decisions with cost-benefit analysis.
Where intervention is needed, government should embrace experimentation and evolution over grand designs. A proportion, around £1 billion of the Department for Work & Pensions’ circa £175 billion budget should be used to fund policy experiments to find better solutions for sustained joblessness. This could test policies like Finland’s proposal for a lifelong learning voucher scheme.
Robot taxes should be rejected. They are poorly conceived, would hinder progress, and would be ineffective in a globalised economy.
A popular policy to protect against the worst AI scenarios is a Universal Basic Income (UBI). The less fashionable Negative Income Tax offers an attractive formulation to achieve this outcome. It would ideally be paired with flatter income taxes.
The Government should complete experiments and to continue to refine welfare and tax policies to build upon the current system of Universal Credit.